General Motors has abandoned its multi-billion dollar outsourcing agreements with HP and other providers in favor of an internal hyperscale environment that runs Hadoop on low-cost x86 servers. This week the carmaker launched its first data center in Warren, Michigan.

The $130 million facility, the first of two, can hold 20,000 square feet worth of commodity servers. That means it can accommodate over 10,000 ‘pizza box’ size machines, smaller-size systems and several larger mainframes, according to GM’s Jeff Liedel. The executive reported that the subfloor is wired with 1,600 miles of fiber cable and more than 400 miles of copper cable, and said that the facility’s two data halls are fitted to accommodate 24 server clusters each. The Warren facility runs standard Oracle software in conjunction with Hadoop. GM has integrated IBM’s InfoSphere Big Insights platform with Teradata’s MPP-based 6700 Enterprise Data Warehouse to create an “advanced, custom data warehouse.”

“What they are focused on is building highly efficient data centers. It is a dominant trend right now,” commented David Cappuccio, Director of Research at research firm Gartner. “In general, these ideas first emerged in hyper scale data centers, like those of Google,” Cappuccio said. “We are seeing more and more companies interested in these ideas.”

GM’s decision to deploy Big Insights is a big win for IBM, which ranked as the the number one Big Data vendor by revenue. But it’s not the only benefit to GM’s internal project. Yesterday, StorageANGLE editor Ryan Cox outlined the perks of GM’s new data center as it pertains to storage costs and streamlining operations, noting the growing trend towards building in efficiency for storage architecture.

Late last month the vendor unveiled MessageSight, an appliance designed to serve as a “Big Data control node” for networks that service millions of devices. SiliconAngle’s Mike Wheatley provided his take on the system in a recent article.